Shopping campaigns are the one place in Google Ads where you cannot add a keyword, and that single fact rewrites how you control traffic. There is no positive keyword list deciding which searches your product ads show for — Google matches queries against your product data feed instead, reading your titles, descriptions, and product types to decide relevance. So the lever you reach for on Search — add the keyword, set the bid — simply is not there. The only keyword-level control Shopping gives you is the negative keyword, which means query hygiene here is entirely subtractive: you shape traffic by blocking what you do not want, never by bidding up what you do.
That makes the search terms report more important in Shopping than in Search, not less, because it is the only window into a matching process you otherwise do not touch. This post treats Shopping as its own query-control problem: why the feed replaces keywords, where to find the search terms report and what Performance Max withholds, the per-query-per-product bid trap that makes some waste impossible to bid away, the patterns worth negating first, and the priority-plus-negatives structure practitioners use to tier traffic. Google's guidance is the anchor throughout — the search terms report is where you find the actual queries that triggered your ads and turn them into negatives, and in Shopping those negatives are the whole toolkit.
Shopping has no positive keywords — the feed matches for you
The defining mechanic of a Shopping campaign is that you never tell Google which searches to match. On a Search campaign you build a keyword list and Google matches queries against it; on Shopping there is no keyword list at all. Eligibility is decided by matching the search against your product data feed — the product title, description, product type, brand, and other attributes you submit through Merchant Center. A query that shares meaning with your product data makes the product eligible; one that does not, does not. Your control over what you match, therefore, lives mostly in feed optimisation, and your control over what you exclude lives entirely in negative keywords.
This inverts the usual mental model. On Search you think first about the keywords you want and treat negatives as cleanup; on Shopping there is nothing to opt into, so negatives are not cleanup, they are the primary steering wheel. Every irrelevant query that reaches your products — the wrong intent, the wrong product category, the informational searcher who will never buy — can only be stopped one way, by adding it or its pattern as a negative. If you have carried over a Search habit of building keyword structure and treating the search terms report as an occasional audit, Shopping demands the opposite discipline: the report is your only feedback loop, and negatives are the only action it produces. The underlying logic is the same one behind keywordless Search ad formats where negatives are the only control — when the positive keyword disappears, the negative list carries the entire weight of query control.
Where the Shopping search terms report is — and what PMax hides
For a Standard Shopping campaign the report is exactly where you expect: open the campaign, go to Insights and reports, then Search terms, and you get the literal queries that triggered your product ads, with clicks, cost, and conversions attached. Sort by cost descending and filter to zero conversions, and the waste sorts itself to the top — the fastest way to see the queries eating budget without paying it back. This is the same workflow as the hidden wasted spend hunt on Search, and the same thresholds apply: a term with real cost and no conversions is a candidate, and the question is only how much spend to tolerate before you act.
Performance Max, which now carries the Shopping inventory for many accounts, is stingier with the same data. It exposes a search terms view, but it aggregates queries into themes and withholds more of the literal detail than Standard Shopping does, so you often see the shape of the traffic rather than every term. That matters because the negatives you can add against it are correspondingly coarser, and the reporting gap is the same one that makes Performance Max search terms harder to negate cleanly. The practical consequence: if you want full visibility and full negative control over your Shopping traffic, Standard Shopping still gives you more of both, and any account running product ads through PMax should treat the search terms it does surface as a sample to act on aggressively rather than a complete list to work through line by line.
One query, two products: why you cannot bid the waste away
The subtlest trap in Shopping is that a single query can be right for one product and wrong for another, and you cannot resolve that with a bid. Bids in Shopping are set per product group, not per query, so when the search "running shoes" makes both a $40 trainer and a $200 specialist shoe eligible, there is no bid that is simultaneously correct for the cheap product it converts and the expensive one it does not. On Search you would split the term into different ad groups with different bids; on Shopping the query is not yours to bid on at all. As Optmyzr notes, a term can be unprofitable for one product but great for another, and while you cannot segment for that manually, Smart Bidding can.
That leaves exactly two honest responses to a query that is bad for some of your catalogue. The first is Smart Bidding: let the algorithm read conversion data query by query and decide, impression by impression, whether this search is worth bidding on for the products it can serve — the only mechanism that reacts at the query level in Shopping. The second is the negative keyword, which removes the query entirely. The mistake to avoid is imagining a third option that does not exist: quietly lowering a bid on the bad term. There is no such control. So the decision is binary and worth making deliberately — either the query has enough legitimate demand across your feed to trust Smart Bidding with it, or it is clean waste and you negate it. This is the Shopping form of the negatives-versus-Smart-Bidding division of labour: the algorithm handles the ambiguous middle, negatives handle the queries that are wrong no matter which product they touch.
The patterns worth negating first in Shopping
Start with intent mismatches, because they are the cleanest waste in a Shopping account. Informational and research queries — "how to", "what is", "reviews", "instructions", "repair", "manual" — trigger product ads constantly and rarely buy, because the person is reading, not shopping. Free-intent and second-hand modifiers are the next tier: "free", "used", "second hand", "cheap", and "DIY" pull in searchers whose intent your feed cannot satisfy profitably. These are broad-negatable patterns rather than one-off terms, so they belong in the same pattern-grouping approach you would use on Search — block the recurring token, not each individual query it appears in.
Then handle category and product-line bleed, which is specific to how the feed matches. Because Google reads your product data broadly, a query for an adjacent product you do not sell — the accessory when you sell the device, the part when you sell the whole, the wrong brand a shared word dragged in — will surface your ad and spend. Read the report for product terms that are close to yours but not what you stock, and negate them so the feed's generosity does not become your leak. Competitor brand names are their own decision here, exactly as on Search: sometimes worth serving, often worth blocking, and always worth a deliberate negate-or-bid call rather than a reflex. Work from the report in every case — the point of Shopping negatives is to block what your feed actually attracted, not what a generic junk list guesses it might.
Priority plus negatives: tiering traffic between campaigns
The most powerful thing negatives do in Shopping is route traffic between campaigns, using the campaign priority setting. Priority — low, medium, or high — decides which of your Shopping campaigns enters the auction when more than one could serve the same product for the same query. On its own, priority would send everything to the highest-priority campaign; negatives are what make it useful. The classic structure runs a high-priority campaign that holds broad, generic, low-value queries at a low bid, with negatives that block your valuable specific terms — and those blocked terms then fall through to a lower-priority campaign that bids up specifically for them.
Concretely: a high-priority "generic" campaign catches "shoes" and "running shoes" cheaply, but has negatives for your exact product and brand queries; a medium- or low-priority campaign, with no such negatives, is the only one left eligible for "nike air zoom pegasus size 10" and bids accordingly. Priority sets the order the campaigns are considered; the negatives do the actual sorting of which query lands where. This is the Shopping analogue of negative fencing to stop broad match cannibalising specific terms — you are using negatives to keep cheap generic traffic and valuable specific traffic in separate buckets so each can be bid correctly. Get the fencing wrong and the tiers collapse into each other, so this is a structure to document and audit, not to set once and forget.
Where Shopping negatives belong across the account
The three levels are identical to Search — ad group, campaign, and account-level shared lists — but the tiering structure above changes how you use them. Universally unwanted queries go on a shared list so every Shopping campaign inherits them: languages you never serve, plainly irrelevant categories, and the research-intent junk that no campaign should ever pay for. This is the same account-versus-campaign-versus-shared list discipline that keeps universal truths high and specifics low, and it keeps your maintenance in one place rather than copied across campaigns you will inevitably forget to update.
The negatives that do the priority-tier routing, by contrast, must stay at the campaign level, and this is the mistake to watch for. If you put the negatives that fence your specific terms out of the generic campaign onto a shared list, they apply everywhere — including the lower-priority campaign whose whole job is to serve those specific terms — and the tiering collapses silently. So keep routing negatives scoped to the campaign that needs them, and reserve the shared list for blocks that are true everywhere. Ad-group negatives have a narrower role: splitting queries between product groups inside a single campaign, the same way you would separate intent within one Search campaign. The rule of thumb is unchanged from Search — ask whether the block should be universal; if yes, lift it high; if it exists to steer, keep it local.
Make it a routine, not a rescue
Because Shopping has no positive keywords, its search terms report drifts faster than a Search campaign's does — there is no keyword structure quietly constraining what you match, so every feed change, new product, and shift in Google's matching can introduce fresh queries overnight. Treat the report as a standing review rather than a rescue you run when spend looks wrong. Pull it on a fixed cadence, sort by cost, filter to zero conversions, and negate the patterns before they accumulate, using the same click-and-cost thresholds you would apply on Search to decide when a term has earned a negative rather than acting on a single expensive click.
Fold the priority-tier audit into the same routine. Confirm your fencing negatives still route traffic the way you intended, because a new product or a widened feed can make a term newly eligible in the wrong tier, and a broken fence leaks the exact valuable queries the structure exists to protect. Run Shopping hygiene as part of a broader search hygiene audit, and hold it to the same wasted-spend thresholds and scorecard you use elsewhere. With the feed doing the matching and negatives doing the steering, Shopping traffic quality becomes a maintained line item instead of a periodic surprise.
Related reading
For the other surface where positive keywords disappear and negatives carry everything, see Performance Max negative keywords. For the division of labour between blocking and letting the algorithm decide, see negatives versus Smart Bidding auto-suppression, and for where the negatives themselves belong, see negative keyword list structure across account, campaign, and shared.